Research On Technologies Of Intrusion Detection For Wireless Sensor Network

Information security is the key to the large-scale application of Wireless Sensor Networks(WSN).Intrusion detection technology can provide a comprehensive and deep level protection of wireless sensor networks.Aiming at the problems of limited energy of the wireless sensor network node,low intrusion detection accuracy and requirements of real-time detection,this paper focuses on the research of intrusion detection model and algorithm and proposes a intrusion detection model based on improved random forest(RF)algorithm for wireless sensor network.The main works of this paper are as follows:1.This paper describes the energy,computing capability and storage capacity of wireless sensor network node are limited.If the intrusion detection system model has high complexity,it will consume too much energy resource in the practical application,which will lead to the decrease of the life cycle.Intrusion detection model should be fit for the features of wireless sensor network and detect intrusion behaviors correctly and timely.2.A hierarchical intrusion detection model is adopted in this paper.Complex analysis and detection functions are realized on the task management node,and the common sensor nodes and the sink nodes realize the function of acquisition and feature extraction respectively which are not required to calculation and storage,in order to reduce the system's overall cost.In view of the lack of real-time performance of Hierarchical Intrusion Detection Model,the improvement of throughput in Bluetooth data transmission is studied,and the realization of Bluetooth adaptive packet selection strategy is analyzed.A signal-to-noise ratio(SNR)estimation method based on power spectrum and threshold is studied.The algorithm can estimate the channel signal-to-noise ratio accurately and in real time,and provide the premise for the realization of Bluetooth adaptive packet selection strategy,so that the improvement of throughput in Bluetooth data transmission can be realized.3.Intrusion detection algorithms in wireless sensor network are deeply studied.Genetic algorithm based selective ensemble(GASEN)algorithm could select the appropriate parts from a set of available neural networks to form a new combination.This research is based on GASEN algorithm and adopts OOB estimation of random forest as validation criteria to evaluate the performance of the set of selected decision trees.The intrusion detection model based on an improved random forest algorithm in wireless sensor network is build.The results of simulation experiments on the KDD Cup 99 data set show that compared with standard random forest algorithm,the proposed algorithm could improve both the real-time and accuracy of intrusion detection.